Entry Pricing and Product Quality: The Case of Antidepressant

Drugs

Jie Chen

Department of Economics Stony Brook University

John A. Rizzo

Department of Economics Stony Brook University

June, 2007

ABSTRACT

The rising prices of pharmaceuticals have generated considerable, and often acrimonious, debate. Yet, there is little theoretical work or empirical evidence on pharmaceutical entry-pricing strategies or on the time paths of the prices of new entrants versus incumbents. This study explores the entry pricing strategies of pharmaceutical manufacturers.

Previous theoretical models have considered a “price skimming” strategy, with drugs entering the market at a premier relative to incumbents or a “market penetration” strategy, pricing the drug at a discount in the hope of gaining market share. The selection of appropriate strategy depends upon the degree of product differentiation and/or the nature of repeat purchase arrangements. However, these investigations have not examined the implications of product quality for entry pricing and pricing dynamics.

The present study provides a model that demonstrates the interplay between quality and product differentiation in determining the optimal pricing strategy. In particular, higher (lower) quality products will engage in price skimming (market penetration) strategies in markets where products are sufficiently differentiated, but will choose a market penetration (price skimming) strategy in markets that are less differentiated. The reason is that in less differentiated markets, price competition is more important than quality competition for gaining market share and higher-quality firms are better able to garner greater market share through price cuts.

We tested this model using a unique database that combines information on drug price and quality for antidepressant drugs during the years 1999-2002. Based on quality perceptions by physicians, the antidepressant market we examine is a relatively homogeneous drug class, which consisted largely of selective serotonin reuptake inhibitors (SSRIs).

The results indicate that higher quality antidepressants engage in a market penetration strategy, charging initially lower prices that rise over time. Five years post-entry, we find that prices of all antidepressant drugs examined tend to converge. These findings are consistent with a market in which product differentiation is modest and consumers learn which drug works best for them through experience.

JEL Classification: I110, L100, L110

1

I. INTRODUCTION

Pharmaceuticals account for a significant share of total health care expenditures in the United States.

In 2003, Americans spent $179.2 billion on prescription drugs, almost 4.5 times the $40.3 billion

spent in 1990. Importantly, the rate of spending growth for prescription medications outpaced other

areas of medical care from 1995 to 2003. During the period from 1995 to 2002, pharmaceutical

manufacturing was the most profitable industry in the United States (Kaiser 2005).

The rapid increase in pharmaceutical spending has generated both interest and controversy over drug

pricing. Yet, there is little theoretical or empirical evidence on entry pricing strategies or the time

paths of the prices of new entrants versus incumbents. This study explores the entry pricing strategies

of pharmaceutical manufacturers. Previous work has suggested that market conditions determine

whether entrants choose lower penetration prices or higher skimming prices. Specifically, in markets

where products are relatively well-differentiated and repeat purchase arrangements are less common,

entrants are thought to adopt a market-skimming entry pricing strategy, charging a relatively high

initial price. However, in more homogeneous markets and in those where repeat purchases are more

common, one might expect a market penetration pricing strategy by an entrant (Eaton and Lipsey

1989; Schmalensee 1982; Dolan and Jeuland 1981; Rao 1984). In such markets, there may be greater

uncertainty as to whether the entrant’s drug is a significant improvement. Thus, the manufacturer may

adopt a low-price entry strategy to familiarize consumers with the product, reaping the benefits in the

form of higher market shares and prices over time.

Product quality has also been thought to affect entry pricing. For instance, Dean (1969) argued that

higher-quality entrants will engage in a skimming strategy, while entrants that offer only marginal

improvements will adopt a market penetration strategy, charging lower initial prices. Previous

2

research seems to suggest that (1) the degree of product differentiation and nature of repeat purchase

arrangements affect the decision to engage in skimming or penetration pricing strategies and (2)

substantially higher-quality products tend to adopt skimming strategies while modestly better products

will engage in market penetration strategies.

While earlier theoretical models considered the roles of product differentiation and repeat purchases

on pricing strategies, existing models have not explicitly derived the role of quality in affecting entry

pricing strategies and subsequent pricing decisions. Moreover, existing studies on the effects of

product differentiation and repeat purchase arrangements on entry pricing are fundamentally across-

market issues, i.e., they account for alternative entry pricing strategies across markets that are

distinguished by their product differentiation and/or the nature of repeat purchases of their products.

However, these studies do not tell us if and when firms within a product market might engage in

skimming or market penetration pricing strategies. Implicitly, these models assume that all firms

within a given market adopt the same entry pricing strategy.

Thus, no current theory suggests how firms within a product market choose to engage in skimming or

market penetration pricing strategies based on differences in product quality. Such a model would

need to derive not only the relationship between quality and the entry price, but also the evolution of

price over time.1 Moreover, no previous work has examined the potentially important interactions

between quality and product differentiation on optimal entry pricing strategies. In particular, no

study has considered whether the relationship between product quality and pricing might vary

depending upon the degree of product differentiation within a given market.

1 For example, to formally derive a skimming strategy, one needs to show conditions for which higher quality

leads to a higher entry price and conditions under which that price declines over time.

3

To help bridge these gaps in the literature, we provide a model in which entry pricing and pricing

dynamics are determined as a function of product quality. Here, we demonstrate that the decision by

manufacturers of higher-quality products to engage in skimming versus market penetration price

strategies itself depends upon the degree of product differentiation within the market. In relatively

homogeneous markets, market share is quite sensitive to price and a market penetration strategy is

beneficial to makers of the higher-quality products. This is consistent with Comanor’s (1986)

assertion that price competition is more important when the market is relatively homogeneous.2 In

such cases, our model shows that market share of the higher-quality products is particularly sensitive

to price, so that, for example, higher-quality drugs will set low initial prices to help generate greater

market share. In more differentiated markets, firms manufacturing higher-quality products will

optimally choose a market skimming strategy.

Our model is tested empirically using a nationally representative data set on drug utilization and

expenditures combined with a physician survey on the quality attributes of drugs to examine the effect

of drug quality on pharmaceutical pricing strategies. Our quality measure consists of a comprehensive

physician assessment of drug attributes and it provides an overall index of a drug’s efficacy and side

effects.

Specifically, we examine the brand name antidepressant drug class during the period 1999 to 2002.

Based on physician quality perceptions, the brand name antidepressant market is a rather

2 Comanor 1986: “…where sellers of new products have not been able to achieve substantial quality advantages,

they rely more on price competition to enter a therapeutic market. Products that embody higher quality, on the

other hand, are more distant from the competitive pressures of established products and can be priced at higher

levels.”

4

homogeneous drug class, consisting largely of selective-serotonin reuptake inhibitors (SSRIs). Using

the data described above, we find strong evidence that antidepressant drugs adopt a market penetration

pricing strategy, that is, higher-quality antidepressants enter at lower prices, but their prices rise over

time. Low-quality antidepressants enter at higher prices and their prices decrease over time. These

results are consistent with our model, namely, that market share is particularly sensitive to price for

antidepressants with relatively high quality. Thus, the optimal strategy for higher-quality

antidepressants is to adopt a market penetration pricing strategy.

Interestingly, we find that five years post-entry, the prices of the antidepressant drugs converge. This

is consistent with consumer-search behavior: consumers switch among the drugs to find the one that

works best for them -- which is not necessarily the one with the highest physician valuation -- and

prices are adjusted over time via the switching process. As the antidepressant market is relatively

homogeneous, the switching process is relatively stable and symmetric and all of the prices converge

after 5 years.

The remainder of this paper is divided into seven parts. Part II summarizes the literature on entry

pricing and empirical work as it pertains to the pharmaceutical industry. The antidepressant drug

market examined in this study is described in Part III and the conceptual framework is presented in

section IV. Part V describes the data and empirical models to be estimated and the results are

presented in Part VI. Part VII summarizes the findings and their implications.

II. PREVIOUS WORK

There are two distinct types of rivals in pharmaceutical markets. First, there is competition from the

branded substitutes, which typically embody different chemical entities and enjoy patent protection,

5

and second, there is competition from generic versions of new and existing products (Rao 1984).3

Unlike different generic versions of the same drug, brand name competitors are differentiated in terms

of quality. As we are particularly interested in the role of quality in entry-pricing strategies, our

analysis focuses on brand name-drugs.

THEORETICAL STUDIES

Dean (1969) distinguishes between two strategies for pricing new products: “skimming” pricing and

“penetration” pricing. Skimming is the strategy of setting a high introductory price and then lowering

it over time, while penetration is the strategy of setting a low price for a new product to gain market

share, raising the price thereafter. Dean suggests that skimming may be used to extract the highest

willingness to pay among consumers. This strategy may also be adopted when a product has few

“close rivals” in its early stages.4 On the other hand, Dean asserts that penetration is more common

3 In much of this literature, price competition is not considered until patents have expired and generic

substitutes enter. For instance, Frank and Salkever (1997) find higher prices following generic entry, while

Caves et al. (1991) find a slight decline in a branded drugs’ price after the appearance of generic rivals.

Grabowski & Vernon (1992) find the price trends of the original brand name drugs following generic entry

differ by drug category. Thus, evidence on the effect of generic competition on the prices charged for the

original brand name drugs is mixed.

4 That is, price skimming will occur in more differentiated markets. Dean discusses these ideas but does not

formalize them in terms of a model. Eaton and Lipsey (1989) make a similar argument in the context of a

spatial model of product differentiation.

6

when market share is very sensitive to the price or when the products face drastic (potential)

competition. This work gives an insightful qualitative analysis, but does not provide a formal model.

A number of theoretical studies of pricing strategies followed Dean’s seminal work. Schmalensee

(1982) considers the riskiness of new products, suggesting that pioneering products should charge low

entry prices, because the manufacturer must persuade customers to try its “ex ante risky product” in

order to build its reputation. In addition, late entrants face substantial disadvantages relative to

pioneering brands, so they need to set even lower introductory prices. Therefore, the optimal strategy

for the later entrants is to differentiate themselves from the predecessor and avoid being perceived as a

“me-too” brand.

Shapiro (1983) explored the optimal pricing strategy of a firm with an experience product, in which

consumers learn about quality through use of the product itself. He states that firms’ pricing strategies

may be categorized according to customers’ initial evaluations of the product: overestimation and

underestimation of the product (i.e. optimistic and pessimistic evaluations). If consumers overestimate

the quality of the product, the firm will milk its reputation. In this case, the optimal strategy is to set a

high introductory price and then decrease the price over time. In the underestimation case, the

product’s reputation must be established and the best strategy in this case is to set a low introductory

price and raise it over time. Shapiro considered the role of consumers’ information, but not product

signaling from the supply side.

Bagwell and Riordan (1991) model entry prices as a signal of quality. At the beginning of the period,

the firm sets a price and then consumers form their beliefs about product quality on the basis of this

signal. This belief is updated each period. As consumers become more informed about the drug’s

7

quality, the price distortion lessens. The investigators argue that the most efficient way for the firm to

signal high quality of new products is to charge high introductory prices.

The studies by Schmalensse (1982), Shapiro (1983) and Bagwell and Riordan (1991) shed light on

how entry pricing may be affected by perceptions of quality (Schmalensee; Shapiro) and by the

impact of price as a quality signal (Bagwell and Riordan). However, these studies do not explore the

means by which products with different quality levels adopt different entry pricing strategies.

Moreover, they do not consider the role of product differentiation on entry pricing.

Dolan and Jeuland (1982) devised the first study to explicitly model drug pricing strategy in an

intertemporal framework. They set up a theoretical model and interpret the skimming and penetration

strategies according to the nature of demand. They conclude that market penetration pricing is optimal

if repeated purchases of products are important and the skimming strategy is optimal if the demand is

stable and production costs decrease over time. However, Dolan and Jeuland do not relate skimming

or penetration strategies to variations in product quality.

EMPIRICAL STUDIES

Reekie (1978) investigated Dean’s entry pricing ideas in the pharmaceutical market by examining the

launching prices for 171 new molecular entities (NMEs) introduced into the U.S market between 1958

and 1975. He finds that the prices of new drugs with important therapeutic gains are significantly

higher than the existing counterparts and that the prices of these drugs decline over time. In contrast,

the prices of imitators are much lower than the existing drugs and these low initial prices are followed

by increases in price over time.

Lu and Comanor (1998) explore the demand-side determinants for New Chemical Entity (NCE) prices.

They examine the pricing strategies of 144 newly patented pharmaceuticals in the United States

8

between 1978 and 1987, finding that the main explanatory elements are the “therapeutic value” of the

product and the competition in the market. Their study shows that the introductory price of drugs

representing “important therapeutic gains” can be two or three times those of existing products.

However, the drugs that largely duplicated existing ones are typically priced at comparable levels.

They also examine the time path of drug prices following entry and find that the prices of the

important new drugs declined by about 13% on average four years after entry into the market, while

the prices of drugs with little or no therapeutic improvement rose by 22% on average.

Following Lu and Comanor’s (1998) approach, Ekelund and Persson (2003) review the pricing

strategies of 246 NCEs in the Swedish market and compare the results with those in the US market.

The Swedish pharmaceutical market is highly regulated, with various forms of price-cap regulations

and other regulatory initiatives to limit pharmaceutical costs. Using the methodology delineated in Lu

and Comanor (1998), they also find that introductory prices are positively correlated with drug quality.

In contrast to the results in the US market, however, they did not find evidence of market penetration

pricing strategy. Instead, all prices decrease substantially over time.

Perhaps due to sample size considerations, empirical studies of entry pricing in the pharmaceutical

industry only consider the effects of quality on pricing across drug product markets. In contrast, we

explore the effects of quality on entry prices within a large drug product market – antidepressants.

This approach allows us to examine whether and to what extent quality affects entry pricing within a

specific drug product class and to examine whether drugs within the same product class adopt

different entry pricing strategies. Before turning to our empirical tests, however, the following section

describes the antidepressant drug product market.

9

III. THE MARKET FOR ANTIDEPRESSANT DRUGS

Depression is one of the most prevalent and debilitating disorders in the United States, having an

estimated lifetime prevalence between 10 and 20 percent (Katon and Sullivan 1990; Kessler et al.

1993; 1994).5 Depression is a chronic illness and patients frequently suffer recurrences or relapse

(Thase 1990). As a result of the high prevalence and chronicity of depression, the market for

antidepressant drugs is quite large. Recent evidence (Drug Benefits Trends 2005) indicates that 10

percent of women and 4 percent of men aged 18 and older currently take antidepressants. The report

also shows that antidepressants were associated with the top 5 highest costs per member per year

(PMPY) in 2002, averaging $50.46.

Antidepressant drug use is common and has increased substantially in recent years. According to

results from the National Ambulatory Medical Care Survey comparing data from 1999 with similar

data from 1985, antidepressant drugs accounted for 13% of the entire increase in pharmaceutical

prescribing (Pomerntz 2003). Sales of antidepressants are quite substantial and are likely to remain so.

Total revenues from sales of antidepressants in the U.S. are estimated to grow from $3.72 billion in

1996 to $14.34 billion by 2006 (Frost and Sullivan 2001 Figure 4-6). SSRIs have dominated these

sales, accounting for nearly 90% of sales in 1996. (Frost and Sullivan 2001 Figure 4-7).

Antidepressant drugs are most commonly associated with the treatment of depression, including major

depression, dysthymia, and depression co-existent with anxiety disorders (Market Measures 2001, p.

III-9). The oldest classes of antidepressant drugs include tricyclics (TCAs) and monoamine oxidase

inhibitors (MAOIs), each of which have been available in the U.S. for decades. While these drugs are

5 Given its high prevalence and chronic nature, the economic burden of depression is very large. Research

indicates that the costs of depression totaled $83.1 billion in 2000 (Greenberg et al. 2003).

10

effective in the treatment of depression, they may have serious side effects. For instance, TCAs may

be lethal when taken in overdose. As a result, these drugs are much less commonly used today.

SSRIs began to appear in the late 1980s and 1990s. They have similar effectiveness to TCAs and

MAOIs, but with better-tolerated side effects. Most recently, additional antidepressants have become

available, including non-selective serotonin reuptake inhibitors (NSRIs) and selective norepinephrine

reuptake inhibitors.

Antidepressants thus include a number of brand name drugs which are only available by prescription.

We studied nine drugs constituting the vast majority of brand-name antidepressant drug sales in the

United States. These drugs are in three therapeutically interchangeable categories of medications:

selective serotonin reuptake inhibitors (SSRIs), which include Prozac (fluoxetine), Zoloft (sertraline),

Paxil (paroxetine), and Celexa (citalopram); selective norepinephrine reuptake inhibitors (SNRIs),

which include Effexor (venlafaxine) and Effexor XR; and “other,” a category that includes Wellbutrin

SR (bupropion), Serzone (nefazodone) and Remeron (mirtazapine).

The antidepressant drugs included in our study are listed in Table 1. The drug introduction dates are

taken from the FDA Orange Book. We also list physician’s average perceived quality for each drug

from the Market Measures (2001) surveys.

[INSERT TABLE 1]

IV. CONCEPTUAL MODEL

TWO-PERIOD MODEL

11

We present a two-period model of a profit-maximizing pharmaceutical manufacturer. The firm

chooses its drug’s initial entry price and the price in the subsequent period. As the marginal

production cost of pills is very low, we may ignore these costs without loss of generality.6

The firm chooses prices for its drug in each time period in order to maximize profits:

(1)

10 ,max

PP )),,,,(),,,,(),,,,((1

),,,,( 111000111

0000 YQQPPmYQQPPQQPPmr

PYQQPPP mmmmmmmm ξξρξαξρ +×+

+

where

0P = the drug’s entry price

1P = the drug price in period 1

mP0 = the average price of all other drugs in the market at time 0.

mP1 = the average price of all the other drugs existing in the market at time 1

Q = the quality of the drug

mQ = the average quality of all other drugs in the market 7

0Y = the number of potential new consumers of the drug in period 0

1Y = the number of potential new consumers of the drug in period 1

),,,,( 00 ξρ mm QPQP = the fraction of potential new customers in period 0 who purchase the drug

10 ≤≤ ρ

6 In the pharmaceutical industry, research and development costs are quite substantial (Dimasi et al. 2003). Such costs are incurred well in advance of product launch and may be regarded as sunk costs at the time that pricing decisions are made. One could include production costs of pills by simply regarding the prices as being net of these small unit production costs without any changes in the model or its conclusions. 7 According to the physician perceptions, the quality of drug does not change significantly over time. In this model, we assume the drug’s own quality and the market quality remain the same in two periods.

12

),,,,( 11 ξmm QQPPm = the fraction of potential customers in period 1 who purchase the drug

10 ≤≤ m

ξ = parameter measuring the degree of market differentiation. 10 ≤≤ ξ 8.

= positive rate of discount r

α = parameter measuring the rate of repeat purchases 10 ≤≤α 9

In period 0, the drug is sold to 0Yρ customers and there are new customers who purchase the

drug in period 1. In addition, some fraction of those customers who bought in period 0 continue to

purchase the drug in period 1 and this fraction will be greater if the drug is used for long-term

treatment ( i.e., it requires more repeat purchases); the expression

1mY

0Ymρα gives the number of these

customers. We assume that 0/ P∂∂ρ < 0 ; Q∂∂ /ρ >0; and < 0. That is, a higher price

reduces the fraction of potential consumers who will purchase the drug while higher quality increases

this fraction. The expression < 0 indicates that a lower entry price leads to greater

market share and that this effect is greater in absolute value (more negative) with increasing product

quality.

QP ∂∂∂ 02 /ρ

QP ∂∂∂ 02 /ρ

10

8 A higher value for ξ indicates greater market differentiation.

9 A market with a high level of repeated purchases has a high level ofα , i.e.α is close to one, meaning that

more customers from pervious period will appear again in the second period because of the long-term

treatment.

10 This may be illustrated with an explicit example:

Let ρ(P, Q) = Q.5P-.5

Then d2ρ/dPdQ = -.25Q-.5P-1.5 < 0.

13

We also observe that the fraction of potential new customers ρ not only depends on the drug’s own

price and quality Q, but also on the market price , market quality , and degree of market

differentiation

0P mP0mQ

ξ . Thus, the pharmaceutical market we envision is a form of monopolistic competition,

in which the firm has to take into account market price and quality conditions when setting its own

price. For purposes of analytic tractability, we take , and mP0mQ ξ as given.

We assume that >0; <0; <0; >0. In other words, a

higher market price will increase the share of consumers purchasing the firm’s drug, but this effect is

smaller at higher levels of that drug’s price. A lower market quality will increase the drug’s demand,

with this negative effect being less at higher levels of that drug’s price.

mP∂∂ /ρ mQ∂∂ /ρ mPP 002 / ∂∂∂ ρ mQP ∂∂∂ 0

2 /ρ

We further assume that and . Thus, if a market is more differentiated,

demand increases less with decreased price and demand increases more with higher quality.

0/ 02 >∂∂∂ ξρ P 0/2 >∂∂∂ ξρ Q

We posit

that there exists a critical value , above which the market can be considered as relatively

differentiated and below which the market is relatively homogeneous.

*ξ

11

Maximizing (1) with respect to and gives: 0P 1P

(2) 0

1

00 1 Pr

mPP

P∂∂

++

∂∂

+ραρρ =0

(3) 11 P

mPm

−=∂∂

Substituting (3) into (2), and total differentiating (2) we obtain the following comparative statics

results:

11 We similarly assume 1/ Pm ∂∂ < 0; >0;mPm 1/ ∂∂ Qm ∂∂ / >0; <0; <

0; >0; <0; and .

mQm ∂∂ / QPm ∂∂∂ 12 /

mQPm ∂∂∂ 12 / mPPm 11

2 / ∂∂∂ 0/ 12 >∂∂∂ ξPm 0/2 >∂∂∂ ξQm

14

(4)

20

21

20

2

00

0

2

1

21

01

221

0

2

00

12

11

PrmP

PP

P

QPPm

rP

PQPm

rP

QPP

QdQdP

∂

∂+

+∂

∂+

∂∂

∂∂∂

∂∂

+−

∂∂

∂∂∂

+−

∂∂∂

+∂∂

−=ραρρ

ραραρρ

By the second-order conditions for a maximum, the denominator in (4) must be < 0.12 Thus, the

denominator for expression (4) is negative and the sign of d /dQ depends upon the sign of the

numerator. We can see that only the first term of the numerator is positive. If the positive effect of

0P

Q∂∂ /ρ dominates, we will observe the skimming strategy, i.e. d /dQ >0. Note, however, that the

three remaining negative terms in the numerator each involve the expression

0P

0/ P∂∂ρ . This expression

will become more (less) important the less (more) differentiated is the market. Thus we can further

interpret the sign of dP0/dQ using market differentiation theory.

(i) Suppose the market is relatively homogeneous: ξ < *ξ

Given this assumption, the negative effect of 0/ P∂∂ρ is more likely to dominate the positive effect

of Q∂∂ /ρ .13 Therefore, d /dQ is more likely to be negative.0P 14

12 The denominator is unambiguously negative if, as seems reasonable, <0. This assumption implies

that

20

2 / P∂∂ ρ

),( 0 QPρ is a concave function of . In this case, when is low, a slight change in will not

influence

0P 0P 0P

ρ significantly. At higher levels of , however, the fraction of new consumers will fall

dramatically with further increases in .

0P

0P

13 Q∂∂ /ρ will be a relatively small positive number given the assumption . 0/2 >∂∂∂ ξρ Q

0/ P∂∂ρ will be a very low negative number given the assumption . 0/ 02 >∂∂∂ ξρ P

Therefore, d /dQ is more likely to be negative. 0P

15

In other words, in a homogeneous market, a high-quality drug is more likely to enter with a lower

initial price because the market share is more sensitive to price for high-quality drugs (recall that

<0). Thus, higher-quality drugs will be particularly successful in achieving significant

market share by choosing a lower entry price. In addition, if dP

QP ∂∂∂ 02 /ρ

0/dQ <0, then lower-quality drugs in

the market will enter at a higher price. For these drugs, market share will not increase as much when

prices are decreased and thus the products may enter the market at a higher price.

(ii) Suppose the market is relatively differentiated, ξ > *ξ

In this case, the positive effect of Q∂∂ /ρ is more likely to dominate the negative effect of 0/ P∂∂ρ

and other terms. 15 Thus, d /dQ is more likely to be positive. 0P

In other words, if the market is sufficiently differentiated, a high-quality drug is more likely to enter

with a higher initial price. Here, the increased quality must generate a large enough increase in the

customer base in period 0 to offset the negative impact of higher prices associated with this customer

base.

In addition, if repeat purchases are uncommon, we would expect α to be low and the discount rate, r,

to be high. Lowerα and higher r would decrease the negative terms in the numerator of expression

1/ Pm14 We may observe that in a homogeneous market, ∂∂ is also a very low negative number, which

confirms our conclusion. In addition, we assume that the third order across differentiations is small enough to

be ignored. For example, . 0/ 03 =∂∂∂∂ ξρ QP

15 Q∂∂ /ρ will be a large positive number given the assumption . 0/2 >∂∂∂ ξρ Q

0/ P∂∂ρ will be a slightly low negative number given the assumption . 0/ 02 >∂∂∂ ξρ P

Therefore, d /dQ is more likely to be positive. 0P

16

(4), raising the likelihood that d /dQ >0. In contrast, higher 0P α and lower r will raise the likelihood

that d /dQ <0. These observations suggest that higher-quality drugs tend to charge lower entry

prices in a homogeneous market and in a marketplace where repeat purchases are more common.

0P

Thus, when markets are relatively homogenous, price competition is more important for gaining

market share and higher-quality drugs enjoy a comparative advantage in gaining market share through

lower entry prices.

How will these entry prices compare to , the average price of drugs already on the market? For

sufficiently high levels of quality, these lower entry prices will be below the prices of drugs already on

the market. In this case, higher-quality drugs will enter at a discount relative to the market. Whether

these entry prices are, in fact, lower than the prices of existing drugs is an empirical issue which we

will examine below. When products are more differentiated, price competition assumes less

importance and the higher quality drugs enter with higher prices. At sufficiently high levels of quality,

the higher entry prices will exceed the existing prices of competitor drugs already in the market.

Again, however, this is an empirical issue.

mP0

The effect of market quality on entry price is given by:

(5)

20

21

20

2

00

0

2

1

21

01

221

0

2

00

12

11

PrmP

PP

P

QPPm

rP

PQPm

rP

QPP

QdQdP mmmm

m

∂

∂+

+∂

∂+

∂∂

∂∂∂

∂∂

+−

∂∂

∂∂∂

+−

∂∂∂

+∂∂

−=ραρρ

ραραρρ

As in expression (4), the denominator for expression (5) is negative and the sign of

depends upon the sign of the numerator. Only the first term of the numerator is negative. In other

words, if the market quality has direct strong effect on the drug’s own demand, we are more likely to

observe <0. Finally, the effect of market price on the firm’s entry price is:

mdQdP /0

mdQdP /0

17

(6)

20

21

20

2

00

00

2

1

21

00

2

00

0

0

12

1

PrmP

PP

P

PPPm

rP

PPP

PdPdP mmm

m

∂

∂+

+∂

∂+

∂∂

∂∂∂

∂∂

+−

∂∂∂

+∂∂

−=ραρρ

ραρρ

The denominator for expression (6) is negative and the sign of depends upon the sign of

the numerator. We can see that only the first term of the numerator is positive. As long as the price

effect on demand dominates cross price effects , we are more likely to observe

<0, and vice versa.

mdPdP 00 /

1/ Pm ∂∂ mP0/ ∂∂ρ

mdPdP 00 /

PRICING DYNAMICS

We may also use our model to study pricing dynamics by comparing the entry and subsequent period

prices of drugs. Again, using the first-order conditions given in equations (2) and (3) and rearranging

terms, we may write:

(7) 01

0

1

0 1εε

εα

−+= rm

PP , where

ρρε 0

00

PdPd

= and mP

dPdm 1

11 =ε =-1 from (2)

Since , 0/ 10 >PP ]0,1[0 −∈ε .

Therefore, we have (8) 11

0 >PP

, when mr

rα

ε+++

−<<−1

11 0

(9) 11

0 <PP

, when mr

rα

ε+++

−>1

10

These conclusions are consistent with demand theory. Demand elasticity 0ε is relatively inelastic (as

in Equation (9)), at the lower part of the demand curve (consider for example the linear demand curve)

18

where price is low and the demand is large. Therefore, is more likely to be less than in this

inelastic range. The opposite is true when

0P 1P

0ε is relatively elastic, as in Equation (8).16

We can rewrite equation (8) in the form: >1, when 10 /PP ))1/(1/(11 0 rm ++−<<− αε

If α is large, r is low, and m is large, then the possible range for 0ε enlarges, which means that

is more likely to be greater than . Conversely, if

0P

1P α is low, r is high, and m is low, then the possible

range for 0ε shrinks and is less likely to be greater than . 0P 1P

SUMMARY

As our comparative static analysis demonstrates, a higher quality product does not necessarily enter

the market at a higher price because pricing also depends upon on market conditions and the nature of

repeat-purchases. Specifically, in a relatively homogeneous market and/or with more repeated

purchases, market share is more sensitive to price for the high-quality drug than it is for the

lower-quality drug. Therefore, the best strategy to gain market share is to set a lower initial price for a

high-quality drug to gain market share. In a more heterogeneous market, market share is not as

sensitive to price. Therefore we may observe the opposite strategy for a high-quality drug, i.e. a

skimming strategy. In other words, price competition is more important when the market is relatively

homogeneous.

These observations suggest that skimming and market penetration strategies depend upon the market

characteristics and the nature of repeat purchase arrangements. The classic price-skimming strategy

16 In the extreme case of α =0, 0ε and 1ε are both equal to -1 from the f.o.c, and and are

indeterminate. In this case, if there are no repeat purchases, a firm will set prices independently in the two

periods.

0P 1P

19

discussed in the literature occurs when a high-quality drug enters the market at a higher price that is

subsequently lowered. Although this certainly may occur in our model, it is also possible that a

high-quality entrant will charge a lower initial price. And, as discussed earlier, prices may rise or fall

over time.

Earlier discussions have focused on two scenarios: classic skimming, in which the entrant charges a

higher price that declines over time, and market penetration, in which a lower entry price is selected

and price rises over time. Our analysis delineates conditions under which four scenarios may occur.

These cases are summarized in Table 2.

[INSERT TABLE 2]

V. DATA AND ESTIMATION

DATA

Our analysis is based on data from the 1999-2002 Medical Expenditure Panel Survey (MEPS)

conducted by the Agency for Healthcare Research and Quality (AHRQ). The MEPS database consists

of a number of files, two of which were employed in our study. The Consolidated File is a person-year

level database, which provides detailed consumer information on health care utilization and

expenditures as well as patients' demographics, socioeconomic characteristics, health, and health

insurance status. The Prescribed Medicines File is an event-level file that includes information on the

utilization and payments for each drug used by survey respondents.17 We converted the Prescribed

Medicines File to the person-year level and then merged it with the Consolidated File for this study.

17 Household respondents provided information on the names of all outpatient medications used by each

household member and the names and locations of the pharmacies where medication was obtained. They were

also asked for permission to request records from these pharmacies. Pharmacy providers were asked to provide

20

Measures of physician-perceived quality indicators of drugs are provided by Market Measures Inc.

(2001), a private medical survey research company. Their survey was conducted among a physician

panel recruited from a random sample of those who treat patients with specific drug product classes.

For example, information on the perceived quality of antidepressant drugs was obtained from a panel

of psychiatrists, internists, and family practitioners who regularly treat patients suffering from

depression and who are thus familiar with alternative antidepressant drugs. The quality evaluation is

the physicians’ perceptions of the performance of the drugs in actual clinical practice, rather than

reports from clinical trials by the manufacturers. Physicians provided rating scores of 1 to 5 for

various attributes of a particular drug, with higher scores representing better quality.

Using this information, we constructed composite quality measures. Physicians were queried about 10

indicators of quality for antidepressants, including patient tolerability and the degree to which each

drug had interactions with other drugs. Physicians rated each dimension of quality on a Likert scale

from 5 (best) to 1 (worst) for all drugs in the class. Complete descriptions of these quality indicators

are provided in Appendix 1. We computed the means of these valuations for each drug as the

aggregate of the QUALITY measure. Therefore, our quality measure consists of the average score of

physician perceptions of drug quality.18

the data necessary to assign a national drug code, which is specific for manufacturers, ingredients, strength,

package size, quantity dispersed, total charge, and sources of payments. The AHRQ performed detailed

matching, imputation, consistency checks, sensitivity checks, and reconciliation algorithms.

18 For the antidepressant data, some indicators are not requested every year. To get a large-enough sample

size, we include the indicators that were generated from 1999 to 2002.

21

The subjects we include in this study are persons who are 18 years and older who had health insurance

during the survey year.19 These criteria left a sample of 5,742 subjects for analysis.

ESTIMATION

We estimate price equations for antidepressant drugs of the form:

(10) εααααα +×++++= MktYearQMktYearQXP 43210)ln(

where P is defined as the average transaction price, i.e., the summation of patient payment and the

insurer payment per prescription of a particular drug for each patient. Previous work on

pharmaceutical pricing has typically used either wholesale prices or invoice prices, neither of which

takes into account the substantial rebates and price discounts often obtained by third-party payers.

(Congressional Budget Office, 1998). Our measure of the transaction price, however, is the true price

received by the manufacturers after such rebates and discounts. All prices are inflation-adjusted to

2000 dollars using the medical care component of the consumer price index.

The variable MKTYEAR indicates how long the drug has been on the market. It is defined as the

difference between the survey year and the drug's approval year. This variable is also a proxy for the

aggregate level of marketing efforts, such as detailing and direct to consumer advertising, which

correlate with the length of time that a drug has been on the market.20

Our quality measure is the average of the physicians’ perception of the quality of each drug, as

19 We elected to exclude uninsured subjects because some of them may receive free drug samples, in which

case our price measures could be biased. In fact, however, including this group produced results (available from

the authors on request) that were quite similar to those in the text.

20 Such marketing efforts tend to be concentrated in the initial years following a drug’s introduction and

typically decline substantially thereafter (Grossman and Shapiro 1984).

22

discussed in the previous section.21 We also include an interaction term between MKTYEAR and

QUALITY to examine whether the relationship between quality and price varies with the length of

time that the drug has been on the market.

The vector X includes other characteristics that could affect the transaction price, such as the health

status, sociodemographic characteristics and drug insurance coverage of the patients/subjects. The

names, descriptions, and summary statistics for variables used in this study are provided in Table 3.

21 We consider the quality measure to be an exogenous variable in our model specification. One might argue

that higher-priced drugs signal higher quality, so that the physician’s perceived quality is to some extent a

function of the drug’s price. However, the evidence repeatedly demonstrates that physicians have very poor

information about drug prices (Allan and Innes 2004; Conti et al. 1998; Glickman et al. 1994; Silcock et al.

1997; Walzak et al. 1994). If true, (1) actual prices should bear little relation to physician perceptions of

prices and (2) physician’s perceptions of price should bear little relationship to their perceptions of a drug’s

quality. Consistent with this evidence, our data indicate the physician’s perception of a drug’s cost is weakly

correlated with the actual transaction price (rho = 0.08). Moreover, the correlation between physicians’

awareness of an antidepressant drug’s cost and their perception of the drug’s quality is very low (rho = -0.02).

In addition, our transaction price measure is at the individual patient level. Even if physicians did have some

knowledge of the average wholesale price, they would have far less knowledge of individual transaction prices,

as our data confirm. Finally, because physicians do not bear the cost of the drug, they have less incentive to

factor cost into their quality assessments.

23

To examine the effects of market quality (MKTQUALITY) and market price (MKTPRICE) on the drug

price, we also estimated the following price equations:

(11)εααααααα

εαααααα+++×++++=

++×++++=iceMktMktQualityMktYearQMktYearQXP

MktQualityMktYearQMktYearQXPPr)ln(

)ln(

6543210

543210

We controlled for heteroskedasticity in the above price regressions using the method proposed by

Greene (1999) and Wooldridge (1999).22

[INSERT TABLE 3]

ANTIDEPRESSANTS REVISITED

Our theoretical model concludes that higher-quality drugs will enter with lower prices in markets that

are not well differentiated and where repeat purchase arrangements are common. These features

characterize the antidepressant drug market well. Thus, while Celexa and Wellbutrin SR have the

highest quality ratings among the antidepressant drugs, their quality advantages are not substantially

superior to other drugs in the group. Hence, it may be more difficult for consumers to perceive real

differences in quality among the antidepressant drugs. In terms of our model, this suggests that we

may be more likely to observe lower entry prices among high-quality drugs in the antidepressant

marketplace.

22 For example, we first obtain the predicted errors from (10), and then regress the square of the predicted error

terms on all the independent variables in (10). The predicted fitted values are the weights we used to correct for

heteroskedasticity. In a small percentage (1%) of cases the fitted values were negative and could not be used as

weights. The results (reported in the text below) were quite similar whether or not we corrected for

heteroskedasctivity.

24

The patterns on scrip use lend further support to these predictions. In particular, Table 1 indicates the

average number of scrips filled per patient for antidepressants (4.70). When repeat purchases are

important, our model predicts that higher-quality drugs will charge a lower entry price. Thus, we

might expect to observe lower entry prices among higher-quality antidepressant drugs. The actual

effects of quality on entry prices, as well as the time path of prices are empirical issues, however, to

which we now turn.

VI. RESULTS

QUALITY AND ENTRY PRICE

Table 4 provides the results of the multivariate price equations. Model 1 includes interactions between

the drug’s quality and market year but does not correct for heteroskedasticity. Models 2-4 correct for

heteroskedasticity. Model 3 adds market quality as an explanatory variable and model 4 adds market

quality and market price.

Table 4 reveals that higher-quality antidepressants enter with lower initial prices, as the coefficient for

QUALITY is negative and highly significant for antidepressants. The positive interaction between

QUALITY and MKTYEAR indicates that, for antidepressants of sufficiently high quality, we will

observe a market penetration strategy, with prices rising over time.

These patterns are consistent with both our model and our understanding of the antidepressant drug

market. In choosing among brand-name antidepressants, a relatively homogeneous market, consumers

may have greater difficulty determining whether new products constitute meaningful improvements.

This factor, plus the greater degree of repeat purchases for antidepressants, suggest that charging a

lower initial price may be a better long-run strategy for high quality entrants into this market.

25

Model 3 shows that market quality has a negative effect on the drug price, but the effect is not

significant. Model 4 shows that market price has a significantly negative effect on the drug price. As

indicated by Equation (6), the effect of market price on a drug’s entry price is ambiguous. Entry price

may be higher when the price of existing drugs is higher—a kind of positive spillover effect. On the

other hand, a higher entry price may cause the firm to forego opportunities to gain market share from

incumbent products by charging a lower entry price. Empirically, we find the latter effect dominates,

so that the firm charges a lower entry price when the market price is higher.

Regardless of whether we control for market price and/or market quality in the model specification,

we find the effect of drug quality on its price is always significantly negative and the magnitude of

this effect changes little across specifications. Thus, the relationship between quality and entry price is

robust.

We also note that the variable HASRXINS is negative and highly significant in the antidepressant

regressions, implying that transaction prices for drugs are lower for persons who have drug benefits.

In the literature, it has been argued (Congressional Budget Office, 1998) that drug manufacturers are

more likely to offer discounts to institutions that have more control over drug distribution channels via

entities such as drug formularies, a common feature of drug benefit plans. Our findings are consistent

with the concept that plans offering drug benefits are able to obtain discounts, leading to lower

transaction prices for the drugs purchased through these plans.

26

PREDICTED PRICES

Using the results from Model 2, we predict the entry price for each drug.23 Specifically, we set the

variable MKTYEAR to zero and estimate the real price of each drug when it entered the market. We

also calculate the prices of the other drugs available at that time. We list these predicted prices in

Table 5.

[INSERT TABLE 5]

Table 5 shows that the entry prices of Celexa and Wellbutrin SR, the two drugs with relatively higher

quality, charged lower prices when they entered the market. Moreover, their prices were lower than

the prices charged for drugs that were on the market when they entered. Thus, Celexa and Wellbutrin

SR appear to have used the “market penetration” pricing strategy. In contrast, antidepressants

associated with lower quality charged higher entry prices when they came into the market, consistent

with the “price skimming” strategy.

Upon market entry, Celexa and Wellbutrin SR had a higher quality than the other available drugs, but

the quality differentials were small in comparison to their competitors. As we hypothesized, when

product differentiation is modest, higher-quality drugs charge lower initial prices, such as occurred

with Celexa and Wellbutrin SR.

23 We use model 2 because we do not have market price and market quality for Prozac, which was the first drug

in the SSRI class. We can obtain predicted values for the other drugs using models 3 or 4, which incorporate

these market effects. These results are quite similar to those reported in the text and are available from the

authors on request.

27

We can also use our model to compute the time paths for the prices of each drug. These are

summarized in Figure 1.

[INSERT FIGURE 1]

Manufacturers of the higher-quality drugs -- Celexa and Wellbutrin SR -- charged lower initial prices

and increased their prices over time, the so-called market penetration pricing strategy. According to

our model, these higher-quality drugs entered the market at a lower price on the theory that they will

be better able to gain market share. In contrast, Figure 1 reveals that lower-quality drugs such as

Remeron and Serzone, entered the market with higher prices that then declined over time. Intuitively,

when lower-quality drugs decrease their price, their gain in market share is less than occurs with

higher-quality drugs. Thus, there is less incentive for the lower-quality drugs to be given a low initial

price.

Figure 1 also reveals that at five years post-entry, the prices of the drugs tend to converge. This pattern

is consistent with a consumer-search model. Recall that our quality measure is the physician’s

perception of the average drug quality. Consumers also observe average drug quality ex-ante, but

ex-ante quality may differ from the consumer’s experience ex-post. The reason is that the effects of a

given drug on any individual are often highly idiosyncratic. In addition, a particular drug may not be

selected because some consumers may have contraindications to that drug even though it offers

relatively high average quality. Other consumers may prefer a drug of slightly less average quality

because it is on the formulary of their drug benefit plan and thus costs them less. Nonetheless, one

would expect that most (but not all) consumers would initially prefer the higher-quality drugs with

lower entry prices. This situation means that most consumers (and their physician agents) initially

purchase drugs according to their average quality ratings, but thereafter will switch among the drugs

28

to find the one that works best for them.24 It is a process of experimentation for an individual

consumer. The drug that works best for an individual is not necessarily the drug has the highest

average quality rating. Nevertheless, one might expect that, over time, relatively fewer consumers

from the higher quality drug group will switch to the lower quality group and relatively more

consumers from the lower quality group will switch to the higher quality group. Therefore, the market

share for the high-quality drug will rise over time, while that of the lower-quality drug will decline.

And, through such a process, it would not be surprising to find that the prices of lower-quality drugs

decline over time while those of the higher-quality drugs rise, precisely the pattern that we observe.

VII. CONCLUSION

In this study, we hypothesize that a drug’s entry price should depend on the interaction between the

therapeutic quality of the new agent and the degree of product differentiation in the market. Our

theoretical model predicts that in markets where products are relatively well-differentiated,

higher-quality entrants will tend to adopt a market-skimming pricing strategy, charging a relatively

high initial price. In more homogenous markets, we expect a high-quality entrant to observe a market

penetration pricing strategy. Markets in which repeat purchase arrangements are common would

reinforce these patterns.

Empirically, we examined brand-name antidepressant drugs during the period 1999 to 2002. On the

basis of physician quality perceptions, product differentiation appeared to be fairly modest. Moreover,

the antidepressant market is associated with substantial repeat purchases of the drugs. Consistent with

24 Although most people treated pharmacologically for major depression and related diseases receive

treatments over an extended period of time, switching among antidepressants is a common phenomenon. See

Woods and Rizzo (1997) for further details.

29

our model, we find that higher-quality entrants engaged in a “market penetration” entry pricing

strategy.

Our results thus indicate that drugs within a given product market may adopt quite different entry

pricing strategies. The optimal pricing strategy depends upon the quality of the drug and the degree of

product differentiation across drugs within that product market. We also find that price differences

across drugs tend to diminish over time. This finding is consistent with the idea that consumers

experiment with the drugs over time and search for the drug that works best. Although our empirical

application focuses on the pharmaceutical industry, our theory may be applied to the study of entry

pricing and pricing dynamics in other product markets as well.

30

Table 1: Antidepressants Summary Statistics

Drug Name Observation Initiated Year Quality # Average Scrips SSRI Celexa 730 1999, Dec. 22 3.73 4.89 Prozac 991 1987, Dec. 29 3.56 5.17 Zoloft 1327 1991, Dec 30 3.6 5.24 Paxil 1330 1992, Dec. 29 3.45 5.15 SNRI Effexor 92 1993, Dec 28. 3.47 3.80 Effexor XR 319 1997, Oct 20. 3.58 5.05 OTHER Remeron 200 1996, Jun 14 3.28 4.54 Serzone 175 1994, Dec 22 3.42 4.30 Wellbutrin SR 578 1996, Oct 4 3.69 4.13 Average Quality: 3.53 Average Scrips: 4.70

Weighted Average Quality: 3.56 Weighted Average Scrips: 4.96

N=5742 Mean Market Year: 8.10

31

Table 2: Alternative Market Pricing Strategies

Classic Skimming Modified Skimming Classic Market Penetration Modified Market Penetration

dP0/dQ>0

0P > 1P

dP0/dQ>0

0P < 1P

dP0/dQ<0

0P < 1P

dP0/dQ<0

0P > 1P

Quality competition dominates

mrrα

ε+++

−<<−1

11 0

Quality competition dominates

mrrα

ε+++

−>1

10

Price competition dominates

mrrα

ε+++

−>1

10

Price competition dominates

mrrα

ε+++

−<<−1

11 0

32

Table 3: Antidepressants: Variable Names, Description, and Summary Statistics (includes all subjects aged 18 years or older who have health insurance) N=5742

Variable Names Description Mean Std Dev

DEPENDENT VARIABLESAVGPAYTOT Average total payment (summation of self and third party

payment) for each drug during a calendar year 80.04 40.64

AGE, GENDER AND MARITAL STATUS

AGE1834 DV=1 if subject’s age is between 18 and 34 else = 0 0.17 0.38

AGE3549 DV=1 if subject’s age is between 35 and 49 else = 0 0.37 0.48 AGE5064 DV=1 if subject’s age is between 50and 64 else = 0 0.27 0.44 AGE6574 DV=1 if subject’s age is between 65 and 74 else = 0 0.10 0.30 AGE75UP DV=1 if subject’s age is 75 and up else = 0 0.08 0.28 FEMALE DV=1 if subject is female else = 0 0.74 0.44 RACE & ETHNICWHITE DV=1 if subject is white else = 0 0.54 0.50 BLACK DV=1 if subject is black else = 0 0.05 0.21 HISPANIC DV=1 if subject is Hispanic else = 0 0.10 0.30 OTHER DV=1 if subject’s ethnic is other else = 0 0.31 0.46 EDUCATION

NOHS DV=1 if subject has less than high school else = 0 0.09 0.28

SOMEHS DV=1 if subject attended HS but did not graduate else = 0 0.16 0.36 HSGRAD DV=1 if subject is HS graduate else = 0 0.31 0.46 SOMECOLL DV=1 if subject attended college but did not graduate else = 0 0.16 0.37 COLLGRAD DV=1 if subject is college graduate else = 0 0.21 0.41 GRADSCHL DV=1 if subject attended graduate school else = 0 0.07 0.25 HEALTH STATUS HEALTHPOOR DV=1 if subject’s health is poor else = 0 0.14 0.34 HEALTHFAIR DV=1 if subject’s health is fair else = 0 0.20 0.40 HEALTHGOOD DV=1 if subject’s health is good else = 0 0.31 0.46 HEALTHVGOOD DV=1 if subject’s health is very good else = 0 0.26 0.44 HEALTHEXC DV=1 if subject’s health is excellent else = 0 0.10 0.30 LOCATIONNORTHEAST DV=1 if subject lives in North East Census Region else = 0 0.17 0.38 MIDWEST DV=1 if subject lives in Midwest Census Region else = 0 0.24 0.43 SOUTH DV=1 if subject lives in South Census Region else = 0 0.39 0.49 WEST DV=1 if subject lives in West Region else = 0 0.20 0.40 YEARYR1999 DV=1 if year is 1999 else = 0 0.18 0.38 YR2000 DV=1 if year is 2000 else = 0 0.19 0.39 YR2001 DV=1 if year is 2001 else = 0 0.30 0.46 YR2002 DV=1 if year is 2002 else = 0 0.33 0.47

33

Variable Names Description Mean Std Dev

HEALTH INSURANCEHASRXINS DV=1 if subject has Rx insurance during all or some of the

period year else = 0 0.66 0.47

QUALITY INDEXQUALITY Average of all the quality indicators 3.56 0.12 YEARS ON MARKET MKTYEAR The Years the drug has been on the market 8.10 3.74

34

Table 4: Brand Name Drug Quality and Real Drug Prices with Dependent Variable : ln( Drug Price)

Model 1 Model 2 Model 3 Model 4

Coef P-Value Coef P-Value Coef P-Value Coef P-Value

QUALITY -0.92 0.00 -0.83 0.00 -0.84 0.00 -0.70 0.00 MKTYEAR -0.47 0.00 -0.43 0.00 -0.42 0.00 -0.40 0.00 QUALITY*MKTYEAR 0.13 0.00 0.12 0.00 0.12 0.00 0.11 0.00 MKTQUALITY -0.16 0.68 -0.36 0.34 MKTPRICE -6.14 0.00 AGE3549 0.05 0.03 -0.05 0.02 -0.05 0.02 -0.06 0.00 AGE5064 0.01 0.78 -0.09 0.00 -0.09 0.00 -0.11 0.00 AGE6574 -0.07 0.03 -0.13 0.00 -0.13 0.00 -0.14 0.00 AGE75UP -0.04 0.20 -0.03 0.17 -0.03 0.17 -0.04 0.08 FEMALE 0.04 0.02 0.05 0.00 0.05 0.00 0.05 0.00 BLACK -0.02 0.59 -0.04 0.24 -0.04 0.24 -0.04 0.16 HISPANIC 0.00 0.94 -0.13 0.00 -0.13 0.00 -0.13 0.00 OTHER -0.02 0.23 -0.03 0.03 -0.04 0.04 -0.05 0.01 SOMEHS -0.06 0.08 -0.09 0.00 -0.09 0.00 -0.08 0.00 HSGRAD -0.05 0.10 -0.16 0.00 -0.16 0.00 -0.16 0.00 SOMECOLL -0.02 0.49 -0.18 0.00 -0.18 0.00 -0.17 0.00 COLLGRAD -0.03 0.40 -0.12 0.00 -0.12 0.00 -0.12 0.00 GRADSCHL 0.05 0.23 -0.04 0.31 -0.04 0.30 -0.03 0.50 HEALTHPOOR -0.02 0.62 -0.01 0.65 -0.01 0.65 -0.01 0.71 HEALTHFAIR -0.03 0.40 0.02 0.45 0.02 0.46 0.04 0.09 HEALTHGOOD -0.04 0.20 -0.08 0.00 -0.08 0.00 -0.08 0.00

HEALTHVGOOD -0.05 0.11 -0.10 0.00 -0.10 0.00 -0.09 0.00

HASRXINS -0.06 0.00 -0.04 0.01 -0.04 0.01 -0.04 0.01 MIDWEST -0.01 0.62 -0.02 0.26 -0.02 0.25 -0.04 0.09 SOUTH -0.01 0.59 -0.12 0.00 -0.12 0.00 -0.13 0.00 WEST -0.01 0.68 -0.06 0.02 -0.06 0.02 -0.07 0.01 _CONS 7.68 0.00 7.56 0.00 8.13 0.00 34.31 0.00

N 5742 5671 5671 5671

ADJ. R SQUARE 0.02 0.14 0.14 0.16

Notes: 1. Drug price is adjusted by CPI Medical Care Component in 2000 dollars 2. All the people in the sample have some type of health insurances and are >=18 years of age. 3. Model 1: OLS estimation 4. Model 2: OLS estimation adjusted the heteroskedasticity. (Some predicted negative value cause 71

observations fewer.) 5. Model 3: OLS estimation with market quality, adjusted heteroskedasticity. 6. Model 4: OLS estimation with market quality and market price, adjusted heteroskedasticity.

35

Table 5: Predicted Entry Real Price of Antidepressants Compared to the Existing Drugs

Drug Name Entry Year Entry Price ($) Quality Existing Drugs

Weighted Average Price of Existing Drugs at Entry Year ($)**

Weighted Average Quality of Existing Drugs at Entry Year ($)**

Skimming Strategy

Paxil 92.Dec 92.74 3.45 82.41 3.54 Prozac 83.49* Zoloft 81.59*

Effexor 93. Dec 90.90 3.47 85.61 3.53 Prozac 83.24* Zoloft 81.76* Paxil 91.26*

Remeron 96. Jun 106.67 3.28 84.75 3.53 Prozac 82.50* Zoloft 82.25* Paxil 86.99* Effexor 87.16* Serzone 90.90

Serzone 94. Dec 94.61 3.42 85.10 3.53 Prozac 82.99* Zoloft 81.92* Paxil 89.82* Effexor 89.64*

Penetration Strategy

Celexa 99.Dec 72.95 3.73 83.08 3.54 Prozac 82.00* Zoloft 82.58* Paxil 84.25* Effexor 84.75* Effexor XR 83.08* Remeron 98.47* Serzone 87.34* Wellbutrin SR 77.93*

Zoloft 91.Dec 81.43 3.60 83.74 3.56 Prozac 83.74*

Effexor XR 97. Oct 83.08 3.58 83.66 3.54 Prozac 82.25* Zoloft 82.41* Paxil 85.61* Effexor 85.95* Remeron 102.49* Serzone 89.10* Wellbutrin SR 76.92*

36

Wellbutrin SR 96. Oct 75.93 3.69 85.35 3.52 Prozac 82.50* Zoloft 82.25* Paxil 86.99* Effexor 87.16* Remeron 106.67* Serzone 90.90*

** Weighted average drug prices are calculated using scrip shares as the weights. * These are individual drug prices Note: 1. The predicted estimates of antidepressants were obtained from the price equation adjusted for heteroskedasticity by

number of scrips:

εααααα ++×+++= XMktYearQMktYearQCPI

Pttttt

t

43210ln

We have:

β̂12.043.083.0ˆln XMktYearQMktYearQP tttt +×+−−=

2. To transform to , we use the smearing estimate. The smearing factor in our study is 1.15. P̂ln P̂

37

Figure 1: Predicted Prices of Antidepressants for Each Drug from 0 to 5 Market Years

Antidepressants Price from Year 0 to Year 5

70

80

90

100

110

0 1 2 3 4 5

Market Year

Rea

l Pric

e

PaxilEffexorRemeronSerzoneCelexaZoloftEffexor XRWellbutrin SRProzac

38

References

Alan, G. Michael and G.D. Innes, “Do Family Physicians Know the Costs of Medical Care?”

Canadian Family Physician 50 (2004), 263-270.

Bagwell, K., and M. H. Riordan, "High and Declining Prices Signal Product Quality," American

Econornic Review 81 (Mar. 1991), 24-239.

Caves, RE., MD Whinston, and MA Hurwitz, “Patent Expiration, Entry, and Competition in the U.S.

Pharmaceutical Industry,” Brookings Papers on Economic Activity: Microeconomics (1991), 3-65.

Cohen, Joel, “Sample Design of the 1996 Medical Expenditure Panel Survey Household Component”

MEPS Methodology Report No.2 AHCPR Pub. No. 97-0027. Rockville, MD: Agency for Health Care

Policy and Research (1997).

Comanor, William S, “The Political Economy of the Pharmaceutical Industry”, Journal of Economic

Literature XXIV (September 1986), 1178-1217.

Congressional Budget Office, “How Increased Competition from Generic Drugs Has Affected Prices

and Returns in the Pharmaceutical Industry”, Washington, DC: Congressional Budget Office (1998).

Conti, G., et al., “Do We Know the Costs of What We Prescribe? A Study on Awareness of the Costs

of Drugs and devices among ICU Staff.,” Intensive Care Medicine 24 (1998), 1194-1198.

Croghan, Thomas W., C. A. Melfi, C. E. William, and A. Chawla, “Cost-Effectiveness of

Antidepressant Medications”, Journal of Mental Health Policy and Economics 1 (1988), 109-117.

Dean, Joel, “Pricing Pioneering Products’, The Journal of Industrial Economics 17 (July 1969),

165-179.

DiMasi JA., RW. Hansen, and HG. Grabowski, “The Price of Innovation: New Estimates of Drug

Development Costs” Journal of Health Economics 22 (2003), 151-185.

39

Dolan, Robert J. and A.P. Jeuland, “Experience Curves and Dynamic Demand Models: Implication

for Optimal Pricing Strategies”, Journal of Marketing 45 (Winter 1981).

Drug Benefit Trends, “Drug Cost Projected to Rise 15% to 16% for Next Five Years”, Trend of the

Month 15 (2003), 7-8.

Drug Benefit Trends, “More Americans Take Prescription Medication”, Trend of the Month 17 (2005),

55.

Eaton, B.C., and R.G. Lipsey, “Product Differentiation”, in R. Schmalensee and R.W. Willig (ed.)

Handbook of Industrial Organization, North Holland, (1989).

Ekelund, Mats and B. Persson, “Pharmaceutical Pricing in a Regulated Market”, The Review of

Economics and Statistics 85 (May 2003), 293-306.

Frank, Richard G., S.H. Busch, and E.R. Berndt, “Measuring Prices and Quantities of Treatment for

Depression”, American Economic Review 88 (May 1998), 106-111.

Frank, Richard G. and Salkever David S., “Generic Entry and the Pricing of Pharmaceuticals”,

Journal of Economics and Management Strategy 6 (Spring 1997), 75-90.

Frost and Sullivan, “U.S. Pain Management Pharmaceuticals Markets” San Jose, CA. (2002).

Glickman, L., E. Bruce, F. Caro, and J. Avorn., “Physicians’ Knowledge of Drug Costs for the

Elderly,” Journal of the American Geriatrics Society 42 (1994), 992-996.

Greenberg P., R. Kessler, H. Birnbaum, S. Leong, S. Lowe, et al., “The Economic Burden of

Depression in the United States: How Did It Change Between 1990 and 2000?” Journal of Clinical

Psychiatry 64 (2003), 1465-1476.

Greene, William H., “Econometric Analysis”, 4th ed. PrenticeHall International, Inc. (July 28, 1999).

40

Grossman, G. M. and C. Shapiro, “Informative Advertising with Differentiated Products,” Review of

Economic Studies 5:1 (1984), 63–81.

Hass, Jennifer S., K.A. Phillips, E. P. Gerstenberger, A. C. Seger, “Potential Savings from

Substituting Generic Drugs for Brand-Name Drugs: Medical Expenditure Panel Survey, 1997-2000”,

Ann Intern Med 142 (2005), 891-897.

Hotelling, Harold, “Stability in Competition,” The Economic Journal 39 (Mar 1939), 153.

Hylan, Timothy R., D. P. Buesching, and G. D. Tollefson, “Health Economic Evaluations of

Antidepressants: a Review”, Depression and Anxiety 7 (1998), 53-64.

Kaiser Family Foundation, “Prescription Drug Trends. Fact sheets 3057-04.”

http://www.kff.org/insurance/upload/3057-04.pdf (November 2005).

Katon W. and Sullivan M., “Depression and chronic medical illness”, Journal of Clinical Psychiatry

51 (1990), 3-11.

Kessler R., McGonagle K., Swartz M., Blazer D., and Nelson C., “Sex and Depression in the National

Comorbidity Survey. I: Lifetime Prevalence, Chronicity and Recurrence.” Journal of Affective

Disorders 29 (1993), 85-96.

Kessler R., et al. “Lifetime and 12-Month Prevalence of DSM-III-R Psychiatric Disorders in the

United States. Results from the National Comorbidity Survey.” Archives of General Surgery 51

(1994), 8-19.

Lu, John Z. and W. Comanor, “Strategic Pricing of New Pharmaceuticals”, Review of Economics and

Statistics 80 (February 1998), 108-118.

Perloff, Jeffrey, V. Suslow, and P. Seguin, “Higher Prices from Entry: Pricing of Brand-Name Drugs”,

Working Paper No. 778, UC Berkeley, (1996).

41

Rao, Vithala, “Pricing Research in Marketing: The State of the Art”, The Journal of Business 57

(1984), 39-64.

Reekie, Duncan W., “Price and Quality Competition in the United States Drug Industry”, The Journal

of Industrial Economics 26 (March 1978), 223-237.

Shapiro, Carl, “Optimal Pricing of Experience Goods”, The Bell Journal of Economics, 14 (Autumn

1983), 497-507.

Silcock J et al. “The Cost of Medicines in the United Kingdom: A Survey of General Practitioners’

Opinions and Knowledge.” Pharmacoeconomics 11 (1997), 56-63.

Schmalensee, Richard, “Product Differentiation Advantages of Pioneering Brands”, The American

Economic Review, 72 (June 1982), 349-365.

Smith W, Sherrill A, “A Pharmacoeconomic Study of the Management of Major Depression: Patients

in a TennCare HMO”, Med Interface 9 (July 1996), 88-92.

Thase M., “Relapse and Recurrence in Unipolar Major Depression: Short-Term and Long-Term

Approaches” Journal of Clinical Psychiatry 51 (1990), 51-57.

Walzak D., S. Swindells, and A. Bhardwaj, “Primary Care Physicians and the Cost of Drugs: A study

of prescribing Practices Based on Recognition and Information Sources,” Journal of Clinical

Pharmacology 34 (1994), 1159-1163.

Wooldbridge J, “Introductory Econometrics: A Modern Approach”, South-Western Pub, (1999).

Wooldbridge J, “Econometric Analysis of Cross Section and Panel Data”, The MIT Press, 1st edition

(October 1, 2001).

Woods S and J. Rizzo, “Cost Effectiveness of Antidepressant Treatment Revisited.” British Journal of

Psychiatry 170 (1997), 257-263.

42

Appendix 1: Quality Indicators of Antidepressants Variable Description Mean Std. Dev. QLTYLF Improve Patient's Quality of Life 4.00 0.28 MLDMOD Effective for Mild to Moderate Depression 4.08 0.25 ENERGY Increased Energy Level/Activation 3.49 0.47 TOLERATE Well-Tolerated by Patients 3.59 0.46 INTERAC Few Interaction with Other Drugs 3.54 0.48 SEXUAL Low Incidence of Sexual Dysfunction 2.86 0.49 CHGWEIGHT No Weight Gain/Change 3.16 0.39 SEDATION Low Incidence of Daytime Sedation 3.43 0.65 AGITATION Minimal Agitation 3.50 0.25 RAPID Rapid Onset of Action 2.95 0.17

43